A new distribution record and updated conservation assessment of the endangered Marañón poison frog, Excidobates mysteriosus (Amphibia: Dendrobatidae)

2020·2.6 MB·English

by  Twomey, E

SALAMANDRA 56(1): 71–74 SALAMANDRA
Correspondence
15 February 2020 ISSN 0036–3375 German Journal of Herpetology
Correspondence
A new distribution record and updated conservation assessment
of the endangered Marañón poison frog, Excidobates mysteriosus
(Amphibia: Dendrobatidae)
Evan Twomey1, Jesse Delia2, Manuel Fashé3, Pablo J. Venegas4 & Lisa M. Schulte5
1) Laboratorio de Sistemática de Vertebrados, Pontificia Universidade Católica do Rio Grande do Sul (PUCRS), Porto Alegre, Brazil
2) Department of Biology, Stanford University, Stanford, California, USA
3) Facultad de Ciencias Biológicas, Universidad Nacional Mayor de San Marcos, Lima, Peru
4) División de Herpetología-Centro de Ornitología y Biodiversidad (CORBIDI), Lima, Peru
5) Department of Wildlife-/Zoo-Animal-Biology and Systematics, Faculty of Biological Sciences, Goethe University Frankfurt,
Frankfurt am Main, Germany
Corresponding author: Evan Twomey, e-mail: [email protected]
Manuscript received: 3 October 2019
Accepted: 28 October 2019 by Stefan Lötters
Excidobates mysteriosus (Myers, 1982) is a species of poi- disturbance. In forest habitats, frogs use both terrestrial
son frog (family Dendrobatidae) endemic to Northern and epiphytic bromeliads (Figs 1C, 1D). In contrast with
Peru and one of the more enigmatic species of the family. rock-wall habitats, low-slope areas that are deforested do
It was described in 1982 from a single specimen originally not contain bromeliads nor frogs. Therefore, E. mysteriosus
collected in 1929 (Myers 1982). This species remained un- seems to require forest cover except in areas where slope is
documented in life until R. Schulte (1990) rediscovered it steep enough to support lithophytic bromeliads. Given its
from near the type locality of Santa Rosa de La Yunga, Ca- restricted range and the fact that much of its habitat is frag-
jamarca department, Peru. Although E. mysteriosus was in- mentary and degraded, E. mysteriosus is currently listed as
itially suggested to belong to the Oophaga histrionica group Endangered on the IUCN Red List of Threatened Species
based on its color pattern, large size (SVL 26–29 mm), and (IUCN 2019). However, Twomey & Brown (2008) identi-
osteology, later molecular phylogenetic work placed it in fied additional suitable habitat for this species using an en-
the genus Excidobates (Twomey & Brown 2008). The spe- vironmental niche model (ENM), mainly localized around
cies is easily recognizable due to its distinctive white “pol- the peripheral slopes of the Huancabamba depression (i.e.,
ka-dot” pattern (Fig. 1A) and has become an emblematic the dry valleys near the confluence of the Utcumbamba,
species of Peru’s exceptionally diverse but threatened am- Chinchipe, and Marañón rivers), raising the possibility
phibian fauna (Jarvis et al. 2015). Here, we report a popu- that this species occurs more widely.
lation of E. mysteriosus from a new locality and provide an During field work in the vicinity of Pedro Ruiz, Am-
updated model of suitable habitat to reevaluate this species’ azonas, in 2010, one of us (MF) found a population of
conservation status. E. mysteriosus occurring ~90 km SE from the type local-
In the roughly 20 years following its rediscovery, E. mys- ity. We visited this site again in 2011 and a single vouch-
teriosus was known only from a handful of sites within er (CORBIDI 10481) was collected and deposited in the
5 km of the type locality. All these sites occur in areas of herpetological collection of Centro de Ornitologia y
heavy anthropogenic disturbance (personal observations). Bio diversidad (CORBIDI), in Lima, Peru. Notably, this
These sites generally fall into two habitat categories: rock record occurs in an area predicted to be suitable habitat
walls and forest (Monsalve-Pasapera 2011). In rock-wall in the previous ENM (Twomey & Brown 2008), raising
habitat, frogs live and breed in large lithophytic bromeli- the possibility that the previous model was accurate and
ads (e.g., Tillandsia cf. ecarinata) growing on the rock fac- that E. mysteriosus occupies, at the very least, a corridor
es (Fig. 1B). All known rock-wall habitats lack forest cover of suitable habitat between the type locality and the new
and thus seem to be relatively immune to anthropogenic locality (Fig. 2E).
© 2020 Deutsche Gesellschaft für Herpetologie und Terrarienkunde e.V. (DGHT), Mannheim, Germany
Open access at http://www.salamandra-journal.com 71
Correspondence
In an effort to improve previous habitat models for threshold = 10; other parameters were set to default val-
E. mysteriosus, we used climate, forest, and topograph- ues. Background sampling included 10,000 points within
ic data and the new locality to model suitable habitat for the area of the climate raster. For climate data we used the
this species. Our overall approach was to classify each of WORLDCLIM version 1.4 dataset for current conditions
the data types to a binary suitable/unsuitable habitat layer, (30 seconds resolution, tiles 23, 24, 33, and 34) (Hij mans
with the final model being a combination of all the binary et al. 2005). Fourteen presence points (all obtained from
layers. We determined the thresholds for binary classifica- our own field work) were used to build the model; due to
tion using data at each occurrence record. However, in all concerns about illegal collection for the pet trade (Brown
cases we tended towards a less conservative estimate (i.e., et al. 2011), these points are withheld but are available on
favoring habitat overprediction) so as not to overly restrict request. The resulting model was converted to a binary
the final model. To determine a suitable climatic niche for raster layer at a threshold of > 0.8 probability of occur-
E. mysteriosus, we ran an ENM with Maxent 3.4.1 (Phil- rence. This threshold was justified in that it included all oc-
lips et al. 2017), under the following conditions: features currence records as well as additional habitat in the region,
= linear, quadratic, product, threshold, hinge; regulariza- but did not include areas of clearly unsuitable habitat such
tion multiplier = 1.0, lq to lqp threshold = 10, linear to lq as the low-elevation arid scrub surrounding Bagua Grande
Figure 1. Excidobates mysteriosus and habitat types: (A) E. mysteriosus from the new locality, adult male, CORBIDI 10481. (B) Rock-
wall habitat near the type locality, Santa Rosa de la Yunga, Cajamarca, Peru. Note abundant lithophytic bromeliads growing on rocks
and lack of forest cover. (C) Forest habitat with terrestrial bromeliads near the type locality. (D) Forest habitat with epiphytic brome-
liads near the type locality.
72
Correspondence
Figure 2. Binary data layers for Excidobates mysteriosus habitat modeling. In panels A–E, suitable habitat shown in dark grey. (A) Suit-
able climate modeled with Maxent (prediction probability > 80%), (B) Suitable elevation (elevation 802–1520 m above sea level),
(C) Suitable forest cover (forest cover > 50%), (D) Suitable slope (slope > 25°), (E) Combination of suitable climate, elevation, and
habitat type. Polygons represent the three general areas discussed in the text: (1) Main “corridor” throughout which E. mysteriosus is
highly likely to be distributed; expanded detail shown in Figure 3; (2) Northern versant of the Cordillera de Colan, occurrence status
unknown; (3) Alto Mayo, occurrence status doubtful. We note that area 1 (main corridor) was drawn to approximate a minimum
convex polygon surrounding the known localities and that there is no geographical barrier between areas 1 and 2. Black scale bar in
each panel = 50 km; red polygons indicate occurrence records; yellow star indicates the type locality.
(Fig. 2A). As the ENM predicted habitats far outside the historically) throughout this range. Other areas of suitable
apparent elevational range of E. mysteriosus, we refined this habitat include the northern versant of the Cordillera de
prediction by determining the observed elevational range Colan (Fig. 2E, area 2) and the Alto Mayo (Fig. 2E, area 3).
of E. mysteriosus and excluding habitats well outside this The Cordillera de Colan area has been poorly surveyed,
range. To do this, we used the Japan Aerospace Explora-
tion Agency (JAXA) ALOS digital surface model (DSM) to
extract elevation data from the occurrence records, result-
ing in an observed elevational range of 902–1420 m above
sea level. We used a 100 m +/- buffer around this range to
define a suitable elevation window for this species of 802–
1520 m (Fig. 2B). We identified forest habitat using the Glo-
bal Forest Change dataset (Hansen et al. 2013); habitat was
deemed suitable if a grid cell indicated > 50% forest cover,
which corresponds to the lower values of forest cover seen
in the known forest sites (Fig. 2C). To identify potential
rock-wall habitats, we calculated a slope raster from the
DSM and defined potentially suitable rock-wall habitats
as slope of > 25° (Fig. 2D). The final model was calculated
from areas of suitable climate and elevation, where either
percent forest cover or slope was also suitable (i.e., defor-
ested, high-slope habitats would be included) (Fig. 2E). All
GIS work was done in QGIS version 3.2 (QGIS Develop-
ment Team, 2019).
Figure 3. Deforestation of the habitat corridor of Excidobates mys-
Our results indicate that there is a continuous corri-
teriosus. Suitable habitat is highlighted; areas within this with <
dor of suitable habitat extending between the type locality
50% forest cover are highlighted in red. Dashed white polygon
and the new locality presented here (Fig. 2E, area 1). Giv- shows boundary of Área de Conservación Privada Copallin (ACP
en that the corridor has occurrence records at either end, Copallin). This map shows roughly the same area depicted by the
it is highly likely that E. mysteriosus occurs (or occurred red box on the overview map in Figure 2.
73
Correspondence
and it is unknown if E. mysteriosus occurs here. This area References
has experienced no appreciable deforestation, being re-
mote and inaccessible (i.e., no roads or navigable rivers Brown, J. L., E. Twomey, A. Amezquita, M. B. De Souza, J. P.
Caldwell, S. Lötters, R. Von May, P. R. Melo-Sampaio, D.
exist in this area) and determining whether E. mysteri-
Mejia-Vargas, P. Perez-Peña, M. Pepper, E. H. Poelman,
osus occurs here should be a conservation priority as it
M. Sanchez-Rodriguez & K. Summers (2011): A taxonom-
would represent a large area of undisturbed habitat. The
ic revision of the Neotropical poison frog genus Ranitomeya
Alto Mayo area has been relatively well surveyed and thus
(Amphibia: Dendrobatidae). – Zootaxa, 3083: 1–120.
far no E. mysteriosus have been found in this area. Addi-
Hansen, M. C., P. V. Potapov, R. Moore, M. Hancher, S. A.
tional habitat southwest from the type locality as well as
Turubanova, A. Tyukavina, D. Thau, S. V. Stehman, S. J.
the slopes south of Bagua Grande were found to have a Goetz, T. R. Loveland, A. Kommareddy, A. Egorov, L.
suitable climate and elevation (Fig. 2A), but lacked forest Chini, C. O. Justice & J. R. G. Townshend (2013): High-res-
cover and/or suitable slope and are therefore unlikely to olution global maps of 21st-century forest cover change. – Sci-
represent viable habitat. ence, 342: 850–853.
Under the assumption that E. mysteriosus is restricted Hijmans, R. J., S. E. Cameron, J. L. Parra, P. G. Jones & A.
to the habitat corridor between the type locality and Pedro Jarvis (2005): Very high resolution interpolated climate sur-
Ruiz (Fig. 2E, area 1; Fig. 3), we calculate the extent of oc- faces for global land areas. – International Journal of Climatol-
currence to be 623 km². Of this, 268 km² (43%) has for- ogy, 25: 1965–1978.
est cover < 50% (Fig. 3, pink shaded area). Given that this IUCN. (2019): The IUCN Red List of Threatened Species, Version
corridor is characterized by forested areas interdigitated 2019-2. – http://www.iucnredlist.org.
with deforested agricultural plots, which are clearly vis- Jarvis, L. E., A. Angulo, A. Catenazzi, R. von May, J. L.
ible from satellite images, low forest cover in this area is Brown, E. Lehr & J. Lewis (2015): A re-assessment of priority
unlikely to be natural and therefore represents deforesta- amphibian species of Peru. – Tropical Conservation Science,
8: 623–645.
tion. However, of this low forest cover habitat, a signifi-
cant fraction (59 km², 22%) occurs in high-slope situations Monsalve, S. (2011): Habitat utilization of the endemic poison
and thus may still represent viable habitat. Overall, tally- dart frog Excidobates mysteriosus in North-western Peru. –
Master Thesis, Linköpings Universitet, Department of Phys-
ing the forested habitats along with the potential rock-wall
ics, Chemistry and Biology, Sweden.
habitats (i.e., low forest cover, high slope) brings the to-
Myers, C. W. (1982): Spotted poison frogs: descriptions of three
tal estimated extent of occurrence within the core habi-
new Dendrobates from Western Amazonia, and resurrection
tat corridor to 414 km². This estimate is substantially less
of a lost species from “Chiriqui”. – American Museum Novi-
than the extent of occurrence currently listed by the IUCN
tates, 2721: 1–23.
(717 km²), but would maintain the current classification
Phillips, S. J., R. P. Anderson, M. Dudík, R. E. Schapire & M.
of E. mysteriosus as Endangered (EN) B1ab(iii), satisfying
E. Blair (2017): Opening the black box: an open-source re-
the following criteria (version 3.1): extent of occurrence <
lease of Maxent. – Ecography, 40: 887–893.
5,000 km², severely fragmented or known to exist at no
QGIS Development Team (2019). QGIS Geographic Information
more than five locations, and continuing decline in the ex-
System. Open Source Geospatial Foundation Project. – http://
tent and/or quality of habitat. In terms of habitat protec-
qgis.org.
tion, a small fragment of the main habitat corridor occurs
Schulte, R. (1990): Redescubrimiento y redefinicion de Dendro-
within the Área de Conservación Privada Copallin (ACP
bates mysteriosus (Myers, 1982) de la Cordillera del Condor. –
Copallin) (Fig. 3); this area should be surveyed for E. mys-
Boletin de Lima, 70: 57–68.
teriosus populations and, if present, may represent a focal
Twomey, E. & J. L. Brown (2008): Spotted poison frogs: rediscov-
point for conservation of this species. Overall, conserva-
ery of a lost species and a new genus (Anura: Dendrobatidae)
tion priorities for E. mysteriosus include (1) additional sur- from northwestern Peru. – Herpetologica, 64: 121–137.
veys within the habitat corridor (e.g., near La Peca) to de-
termine whether E. mysteriosus has a continuous distribu-
tion through this area and occurs within the ACP Copal-
lin, (2) surveys along the north slope of the Cordillera de
Colan to determine the presence/absence of E. mysteriosus
in this region, and (3) expanding the extent of protected
habitat along the main habitat corridor between the type
locality and the Pedro Ruiz locality.
Acknowledgments
We thank Cesar Lopez and Napoleon Monsalve for help in
the field. The specimen (CORBIDI 10481) was collected under
permit RD#0581-2011-AG-DGFFS-DGEFFS issued by the Peru-
vian government.
74